TestSavant.AI icon

TestSavant.AI

Guardrails prompts and outputs using TestSavant.AI.

Scan → Validate

AI-generated

Summary

Validates a prompt or prompt-plus-output using the TestSavant.AI Guard API and routes each item into valid or not‑valid output streams.

Inputs

  • main (required) — Input items; each may contain prompt and/or output JSON fields.

Output shape

An array of two item arrays: [validItems, notValidItems]. Each item has json { valid: boolean, prompt: string, output: string | null, result: any } and pairedItem { item: number }.

Valid items (is_valid === true) go to the first array; items that fail validation or error go to the second.

Examples

Example 1: Input scan with prompt provided

Set scanType to 'input', provide prompt, optionally select scanners or use project defaults.

Example 2: Output scan with prompt and output provided

Set scanType to 'output', provide both prompt and output text, optionally select scanners.

Discussion